BM280 stability

[Johanan]

Hello,

I am trying to measure the pressure inside a box (about 1x1x1 meters), which has a few fans taking air out.

The measured range is 10 to 50 pa. I use 2 BMP280 devices, one outside the box, one inside the box.

I calibrate the difference when no fan is on, I get 0 pa.

Whoever after some time, there is a drift, which I cant explain. I read the data sheet many times, and it seems that I should be able to get an accurate measurement, but I have seen drifts of 30 pa difference with no fan working on a stable system.

I am using "Ultra high resolution" prs oversampling x16, Temp. Over sampling x2, IIR coefficient x16, and I even add my own IIR of x32 is my software. Sample delay is 0.5ms , as power consumption is not an issue.

Stile, I have this drift.

What am I missing? Is there some particular PCB layout needed (I followed all recommendations on data sheet, my PCB has only 3 mounting holes and so on.

Thanks

Johanan

[hor63]

IMHO you have chosen a bad design from the start.

the pressure difference is at the lower end of the resolution of the absolute pressure sensors.

You should use a differential pressure sensor with an appropriate rage ( e.g. 100PA). Here your pressure difference is the full range of the sensor, and not a minuscule part like in your design.

[Johanan]

Well, you are correct that I am starching it, but I am still within the data sheet specifications.

according to my settings, in ultra high resolution, and IIR cof. I set, the nois in pressur measurment shuld be 0.2 pa.

I  avarge this more by my own IIR, and still get fluctuations  3 pa, and slow drifts of 6 - 8 pa in a short time.

So I wonder if the data sheet is inaccurate, or I am doing something wrong, or I misunderstand the data sheet.

Price of differential sensors for this range is much higher then BMP280 or BMP388.

[o_o]

I think you indeed misinterpreted the datasheet. This is short term drift, which is greatly improved in BMP388.

Noise is not the only important parameter.

o_o

[Johanan]

It is true that this was a test project, which I wanted to see if it can be done.

Unfortunately the BMP388 pin out is different, but I will be happy to edit the pcb. Do you think the 388 can do a much better job in such application?

Thanks.